Server load forecaster for capacity planning

The brief

A platform team adds servers by gut feel and still gets surprised on launch days. Forecast traffic and CPU load hour by hour, with weekly patterns and special days, so they can plan capacity a week ahead. Good looks like: a forecast that beats a seasonal baseline, back-tested over several weeks, with a chart of expected peaks.

What you will build

1. Prepare hourly series of traffic or CPU use
2. Build a seasonal-naive baseline
3. Train a gradient-boosted forecast with lag and calendar features
4. Back-test over several weeks
5. Plot forecasts with expected peaks and suggested capacity

Free, open data to build it with

• Alibaba cluster trace ↗ Alibaba (GitHub)
• Web traffic time series forecasting ↗ Kaggle

Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.